Mapping sicknesses in the Netherlands using Twitter

• (Geo)Objective

• Twitter data collection

• Difficulties

• Output

• Lessons learned

• Show how easy it is to collect and manipulate data from Twitter

• Subject with spatial and temporal distribution

• Our search keywords: cold, flu, fever

• (Hopefully) see a nice spatial and temporal trend

• twitteR package

• Twitter API: Rest and Streaming API

• Database in Gaia since November 11 2013

• Filtering with SQL statements

• Commas in tweet as well as .CSV

• Plotting using ggPlot

• Using loop to save files as .PNG

Filtering to remove irrelevant tweets

matrix <- as.matrix(df)
matrix2 <- tolower(matrix)

## Iterate each row in the .CSV file in reversed order, delete the row when a keyword is found
for (row in nrow(matrix): 1){
  line <- matrix2[row, ]
  for (keyword in list){
    if (grepl(keyword, matrix2[row, 1])){
      matrix <- matrix[-row, ]
      break
    }
  }
}

• Regexp not always needed

• Keep it simple

• 'Easy' things aren't always so easy!

• API's make data collection and maipulation accessible for practically everyone